A wind-noise suppressor with SNR based wind-noise detection and speech-wind discrimination is proposed. Wind-noise detection is performed in each frame and frequency based on the power ratio of the noisy speech and an estimated stationary noise. The detection result is modified by speech presence likelihood representing spectral smoothness to eliminate speech components. To suppress wind noise with little speech distortion, spectral gains are made smaller in the frame and the frequency where wind-noise is detected. Subjective evaluation results show that the 5-grade MOS for the proposed wind-noise suppressor reaches 3.4 and is 0.56 higher than that by a conventional noise suppressor with a statistically significant difference.
Masanori KATO
NEC Corporation
Akihiko SUGIYAMA
NEC Corporation,Tokyo Metropolitan University
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Copy
Masanori KATO, Akihiko SUGIYAMA, "A Wind-Noise Suppressor with SNR Based Wind-Noise Detection and Speech-Wind Discrimination" in IEICE TRANSACTIONS on Fundamentals,
vol. E101-A, no. 10, pp. 1638-1645, October 2018, doi: 10.1587/transfun.E101.A.1638.
Abstract: A wind-noise suppressor with SNR based wind-noise detection and speech-wind discrimination is proposed. Wind-noise detection is performed in each frame and frequency based on the power ratio of the noisy speech and an estimated stationary noise. The detection result is modified by speech presence likelihood representing spectral smoothness to eliminate speech components. To suppress wind noise with little speech distortion, spectral gains are made smaller in the frame and the frequency where wind-noise is detected. Subjective evaluation results show that the 5-grade MOS for the proposed wind-noise suppressor reaches 3.4 and is 0.56 higher than that by a conventional noise suppressor with a statistically significant difference.
URL: https://global.ieice.org/en_transactions/fundamentals/10.1587/transfun.E101.A.1638/_p
Copy
@ARTICLE{e101-a_10_1638,
author={Masanori KATO, Akihiko SUGIYAMA, },
journal={IEICE TRANSACTIONS on Fundamentals},
title={A Wind-Noise Suppressor with SNR Based Wind-Noise Detection and Speech-Wind Discrimination},
year={2018},
volume={E101-A},
number={10},
pages={1638-1645},
abstract={A wind-noise suppressor with SNR based wind-noise detection and speech-wind discrimination is proposed. Wind-noise detection is performed in each frame and frequency based on the power ratio of the noisy speech and an estimated stationary noise. The detection result is modified by speech presence likelihood representing spectral smoothness to eliminate speech components. To suppress wind noise with little speech distortion, spectral gains are made smaller in the frame and the frequency where wind-noise is detected. Subjective evaluation results show that the 5-grade MOS for the proposed wind-noise suppressor reaches 3.4 and is 0.56 higher than that by a conventional noise suppressor with a statistically significant difference.},
keywords={},
doi={10.1587/transfun.E101.A.1638},
ISSN={1745-1337},
month={October},}
Copy
TY - JOUR
TI - A Wind-Noise Suppressor with SNR Based Wind-Noise Detection and Speech-Wind Discrimination
T2 - IEICE TRANSACTIONS on Fundamentals
SP - 1638
EP - 1645
AU - Masanori KATO
AU - Akihiko SUGIYAMA
PY - 2018
DO - 10.1587/transfun.E101.A.1638
JO - IEICE TRANSACTIONS on Fundamentals
SN - 1745-1337
VL - E101-A
IS - 10
JA - IEICE TRANSACTIONS on Fundamentals
Y1 - October 2018
AB - A wind-noise suppressor with SNR based wind-noise detection and speech-wind discrimination is proposed. Wind-noise detection is performed in each frame and frequency based on the power ratio of the noisy speech and an estimated stationary noise. The detection result is modified by speech presence likelihood representing spectral smoothness to eliminate speech components. To suppress wind noise with little speech distortion, spectral gains are made smaller in the frame and the frequency where wind-noise is detected. Subjective evaluation results show that the 5-grade MOS for the proposed wind-noise suppressor reaches 3.4 and is 0.56 higher than that by a conventional noise suppressor with a statistically significant difference.
ER -